Re: Welcome to the community group on Bridging GraphQL and RDF

Hi! I'm a data architect at Ontotext. I've been creating semantic models
for the last 9 years, and more and more moving away from heavy ontologies
and towards lighter weight app profiles and rdf shapes.

We released the Ontotext platform that includes GraphQL over RDF at the end
if 2019. I designed the schema language it uses:
http://platform.ontotext.com/soml/index.html. It's currently very simple
and will continue to evolve.

I'm strongly interested if the CG can come up with some best practices for
using GQL over RDF, idiomatic ways to fix GQL shortcomings (of which there
are many (*), iron out details of JSONLD output (which is non trivial),
ways to do multi store federation, etc.

(*) eg
1. To make a subclass, you need to make two things: a type and an interface
2. No multiple types
3. Covariance of object types (data) but not input types (operations)
4. If a mandatory field is missing, the whole object is supposed to bomb
out. Then recursively upward...
5. Prop characteristics can vary per object, which complicates jsonld
output.
6. If a filter var is unset, that can easily break the query.
7. If object id's (IRIs) are missing in a query, one cannot output them,
which is bad for jsonld.

I also hope the CG can define some mappings between schema languages, in
particular SHACL and SHEX, because I think shapes are an important piece of
the equation.
Ruben, where is that gdoc where we gave extra input?

Ps: Check out https://www.slideshare.net/joshsh  by a guy working on the
Uber KG and a main Gremlin contributor, who has good ideas around LPG, RDF,
GQL, HypergraphDB, etc. Few people like Category Theory. However, I think
this paper is worth reading https://arxiv.org/pdf/1909.04881.pdf

Cheers!